SCS · Research notes · Updated 30 September 2026
CS2 Trust Factor: what you can actually check.
Valve shows Trust Factor to nobody, and no public field carries it. What is public is the profile it is built on: how old the account is, what it has actually done in the community, and what Steam has on record. We read that side, score it against every account in our pool, and show you the study the reading comes from, so you can decide whether to believe it.
About a second. Five public sources, no login, cached for 24 hours.
1 free scan a day without an account · results cached for 24 hours
Updated +1,198,295 accounts in the last 7 days
- 25,129,120
- Steam accounts in the pool
- 25,124,706
- checked for bans
- 5,512,853
- with a ban on record · 21.94% of checked
- 29,964,869
- friend-graph edges
Counted inside our pool, not across all of Steam. The pool grows from Steam groups and from crawling the friend lists of banned accounts, so it is not a random sample, and every percentage on this page describes it rather than Steam as a whole.
We do not read Trust Factor. Nobody can.
Valve keeps that number hidden from players and from us alike, and no public field carries it. Any tool that hands you a Trust Factor is showing you its own guess. What we score is what your profile actually shows, and when too little of it is visible we return no score instead of inventing one.
Trust Factor, question by question
10 guides written from the same study as this score
- What Trust Factor is, and what Valve has actually said about it
- Can anyone check Trust Factor? What the checkers really read
- How to improve Trust Factor: what Valve confirms, what we measured
- Red Trust Factor: what actually works, and what only sounds like it does
- New Steam account, low Trust Factor: what the study shows
- Do commends move Trust Factor? What we can measure and what we cannot
- Reports, ban waves and Trust Factor: how the pipeline behaves
- Does account age or Steam level affect Trust Factor?
- Does queuing with a low Trust Factor friend hurt yours?
- The yellow “Trust Factor lower than yours” warning: what it says
01 · What is the score measured against?
the pool: accounts, bans and the friend graph
Updated +1,198,295 accounts in the last 7 days
Also in the pool: 922,895 crawl seed accounts · 26,437 accounts described in full depth · 74 signals checked per scan.
02 · How often do ban waves hit?
bans per day inside the pool
When Valve strikes, the pool shows it. Across the last 180 days the busiest day carried 11,565 bans against a daily median of 199 for those days. Going further back, the all-time daily median is 353, and 8 days sit above ten times that, the heaviest being 2020-08-13 with 58,623 bans, 166 times the all-time median. Counted inside our pool of 25,129,120 accounts, not across all of Steam. Orange marks days above three times the median of the window. Bar heights use a square-root scale so the ordinary days stay readable next to the spikes. 2
03 · Which accounts get banned?
when they were made, and who hides the profile
Account creation years across the pool, 2003 to 2026. The heaviest year is 2015 with 2,500,750 accounts. Creation dates are readable only on public profiles, so this shape describes the public part of the pool.
A hidden profile puts you in the 4.19%. When we cannot measure something, we say so instead of guessing a score. 1
04 · How do we score without reading Trust Factor?
observed patterns, not a Valve metric
2,854 accounts snapshotted while clean that later caught a ban. The decisive threshold is 3,000 cases. Until we reach it, findings stay labelled as observed patterns rather than predictions.
Every feature has to date from before the ban. Without that rule a model learns the aftermath of a ban rather than the signs that came first.
Samples are drawn systematically across the pool. A plain LIMIT N query measures insertion order rather than the population, and it once handed us a sample that was off by five percentage points.
Every percentage we publish carries a Wilson confidence interval. Small samples get wide intervals, and we show the width instead of hiding it.
05 · What moves the score?
four axes and a ceiling
Age, level, badges, library depth, account standing.
Comments, reviews, workshop, groups, identity, look. Pace matters more than volume.
Hour pacing and play patterns between scans. Needs a second scan after 7 to 14 days.
Ban records and banned friends. Facts read from Steam, never our opinion.
The less of a profile we can read, the lower the highest score we are willing to give it. That is not a punishment, it is the limit of what we are prepared to claim about something we can only partly see. A full hundred would mean we checked everything and found nothing wrong, so it has to be earned by there being something to check. Open up what is hidden and the ceiling moves, which you can verify with a second scan.
06 · What is free, and what is in the full report?
the full report is 4.99 EUR, once per profile
Payments are not switched on yet, so today everything in the free column is all there is. Nothing is for sale on this page.
07 · How does a scan work?
five requests, about one second
No login, no password, nothing connected to your account. We read the same public pages any player can open.
The profile page, its XML, ban records, the game library and the friend list. Nothing is written to your account and nothing is changed.
Deterministically. The same profile in the same state gives the same number, every time, and each result records which version of the weights produced it.
Anything hidden lowers the ceiling rather than the score, because a missing measurement is not a bad measurement.
08 · Which Trust Factor myths survive the data?
what you will read on forums, and what we can actually say
A checker can tell you your trust is green, yellow or red.
Nobody can read Trust Factor, ourselves included. Valve keeps it hidden and no public field carries it. Any tool handing you a colour is showing you its own guess wearing Valve's clothes. We score what your profile actually shows, and when too little of it is visible we return no score at all rather than invent one.
Every fresh account starts at the bottom.
A new account does not have to start low. What gets a new account treated badly is jumping straight into competitive with nothing else on the profile, because that shape is shared with throwaway accounts. Age is one of the signals we check, not the whole picture.
Trust resets on a clock, so wait fourteen days and start over.
There is no public evidence of a fixed reset clock, and we are not going to invent one. Honestly, the counter-claim is no better evidenced than the myth: nobody outside Valve has shown that waiting works, and nobody has shown that it cannot. What we can measure is the other side of it, on your own profile: signals that arrive gradually hold their value and bursts do not. So the two weeks are not the point. What you do with them is.
Buy an old, well-stocked account and you inherit its good standing.
Valve answered this one itself, and the answer runs the other way: an acquired account carries whatever standing it already had, and acquired accounts commonly sit very low. That is Valve's own wording, not our inference. Buying one also breaks the Steam Subscriber Agreement, which means the seller can pull it back through support and you have no argument. An aged profile is worth something because of what happened on it over the years, and none of that history transfers to the person who paid for it.
Put three or four hundred dollars into Steam and your standing flips.
The price moves every time somebody repeats it: five dollars in one thread, three hundred in the next, five hundred in the one after. Three numbers that cannot all be true, and they cluster on sites that sell skins and cases — the people who profit from you believing it. Valve has never said spending is weighed at all. What money verifiably buys is Prime, which is a phone number and one purchase, not a sliding scale you can climb.
The number in Looking to Play tells you which tier you landed in.
It tells you how many people are queuing near you at that moment, which moves with your region, the hour and the maps you picked. Even the guides pushing this trick hedge it in the same paragraph. Valve states plainly that the number itself cannot be seen from outside, and a lobby counter is not a back door into it.
Report enough players and you drag your own standing down.
Nothing public supports it, and the forum threads that repeat it contain people saying the opposite two replies down. Our own read — ours to argue, not Valve's to confirm — is duller and more likely: someone who reports every lobby makes their own reports count for less, because a report from a person who suspects everyone carries less information. Reporting an actual cheater costs you nothing.
Expensive skins buy you a good standing.
Inventory value is one signal out of the seventy four we check, and quality beats volume everywhere we score. A profile carrying an expensive inventory and nothing else still reads as thin, both to us and to the people who open it before a match.
1 The pool grows from Steam groups and from crawling the friend lists of banned accounts, so it is not a random sample of Steam. Every percentage on this page describes our pool and nothing wider.
2 Ban dates are derived from the days-since-last-ban field rather than read directly, so they are accurate to about one day. That field measures the distance to the most recent ban, which means older waves are systematically undercounted.
3 Everything here is retrospective. It describes how accounts that were banned tend to look, not that an account which looks that way will be banned. A ban on your report is a fact we read from Steam, never a forecast we made.
What we read from public profiles, why, and how to say no →
Steam Profile Checker v1.6.0 · What's new →

